Created attachment 177029 [details] shar archive passes pourdiere.
Created attachment 177030 [details] shar archive
Created attachment 177033 [details] shar archive
Created attachment 177036 [details] shar archive
A commit references this bug: Author: danfe Date: Thu Jan 19 11:37:45 UTC 2017 New revision: 431857 URL: https://svnweb.freebsd.org/changeset/ports/431857 Log: Add a port of Intel's compiler for high-performance SIMD programming. Because GitHub releases (tarballs) are not fetched with correct modification time, set TIMESTAMP to 1467976432 which corresponds to commit 87d0c9a tagged as this release. PR: 214543 Submitted by: Yuri Victorovich Changes: head/devel/Makefile head/devel/ispc/ head/devel/ispc/Makefile head/devel/ispc/distinfo head/devel/ispc/files/ head/devel/ispc/files/patch-Makefile head/devel/ispc/files/patch-examples_common.mk head/devel/ispc/files/patch-examples_tasksys.cpp head/devel/ispc/pkg-descr
Committed with modifications as of ports r431857, thanks! I've had to fix a few notable bugs, for your consideration: - We typically don't depend on ``bison:devel/bison'' explicitly, there's USES+=bison for that - Entire `post-patch' target is bogus and hardly readable: some variables are better to be passed via MAKE_ARGS (this is a more or less standard practice), and other changes are more maintainable as a patch; notice that this also applies to ${LOCALBASE} since its value is propagated down the make(1) calls: you don't need to "fix it" prior to the build - Ironically, fixing `util.cpp' is easier to do with sed(1) rather than a patch; also notice that deleting bogus #include <alloca.h> is sufficient; <stdlib.h> is already included in that file few lines later - Unless required (e.g., because they get in the way somehow), do not suppress `.bak' suffix when doing in-place sed(1)ing on files, this makes debugging hard - Standard way to fix ${WRKSRC}/*.py scripts is via USES+=shebangfix - Probably couple of other minor issues I forgot ;-)